<<

BRITisH 7 October 1967 MEDICAL JOURNAL 13

Mortality of Bereavement Br Med J: first published as 10.1136/bmj.4.5570.13 on 7 October 1967. Downloaded from

W. DEWI REES,* M.B., B.S., M.R.C.G.P.; SYLVIA G. LUTKINSt B.SC.

Brit. med. J., 1967, 4, 13-16

"He first deceased; she for a little tried control group the assessment was made in each case from the To live without him; liked it not, and died" time of of the individual with whom the control had -Sir Henry Wootton. been paired. That bereavement is associated with an increased mortality rate for close relatives is an old concept that has only recently begun Results to receive statistical confirmation. A total of 5,184 patients lived in the defined survey area; In 1959 Kraus and Lilienfeld reported that married people their age and sex distribution is given in Table I. During the had a lower death rate at every age group than those who were period under review 488 residents died; 404 residents died widowed, single, or divorced and that there was a particularly within the survey area and 84 other residents died in hospitals high excess risk of mortality for young widowed people. Young or private homes outside the area ; the age and sex distribution et al. (1963) then tried to assess the " duration effect " of of these people is given in Table II. Of the 488 residents who widowhood on the mortality of 4,486 widowers aged 55 and died, 108 had no close relatives living in the area and they, over and found that widowhood increased the mortality rate by together with the nine recorded as neonatal , were about 40% in the six months following bereavement. Cox and excluded from the survey. The remaining 371 (76%) form Ford (1964) analysed data about 60,000 women awarded the survey group; the age and sex distribution of this group widows' pensions in 1927 and concluded that the mortality of is given in Table III. When the 371 residents forming the widows is probably not exactly the same in each of the five survey group died they had a total of 903 close relatives living years following bereavement and seems to be higher in the in the area; the comparative figure for the control group is second year after bereavement. The need for further studies 878. The number and type of close relatives extant at the in this field was stressed by Kraus and Lilienfeld (1959) and by time of real or hypothetical death for both the survey and Young et al. (1963), and a survey was therefore made to assess control groups are given in Table IV. The number of each type the effect of bereavement on the mortality of close relatives in a of relative who died in the years following bereavement is small semirural communiy. This survey differed from those previously made in that it included close blood relatives as well TABLE I.-Age/Sex Distribution of Patients Resident Within Survey as the widowed and because a total community study was made Area in an area where the investigator (W.D.R.) possessed a close Age Male Female Total personal knowledge of the people included in the survey.

0-9 375 358 733 http://www.bmj.com/ 10-19 349 303 652 20-29 296 284 580 30-39 329 297 626 Method 40-49 313 354 667 50-59 345 357 702 60-69 288 367 655 The survey was made in a defined area, centred upon the 70-79 157 228 385 80-89 65 101 166 small market town of Llanidloes (pop. 2,350). The area was 90-99 5 13 18 so defined that with very few exceptions all residents were Total 2,522 2,662 5,184 patients of a single group practice. A small cottage hospital,

a chronic sick unit, and an old people's residential home are on 29 September 2021 by guest. Protected copyright. situated within the area. The name, sex, age, address, certified , and place of death of all those people normally resident in the area who died in the six-year period following 1 January 1960 were obtained from the county medical officer of health. The marital state of these people and the identity of those of their close relatives who also lived in the area were determined with the help of the practice secretaries and the secretary of the local cottage hospital. Those people were considered close relatives who were either a spouse, parent, child, or sibling. Those people who had no close relatives alive when they died were excluded from the survey. The remainder were matched by age, sex, and marital state with a similar control group known TABLE III.-Age/Sex Distribution of Deceased Patients (Excluding to have relatives living in the area. This was done by arbitrarily Neonates) With Close Relatives in Area selecting alternate names from the relevant age/sex lists of the practice age/sex register. It was not found possible to match Age Male Female Total all the paired couples exactly for age or marital state, and five 0-91 1 2 10-19 3 - 3 widowed people were paired with people still married. The 20-29 2 - 2 identity of the close relatives of the control group was deter- 30-39 3 2 40-49 6 4 10 mined in a similar way for that of the survey group. In 50-59 27 7 34 60-69 43 32 75 determining the subsequent mortality of close relatives in this 70-79 57 57 114 80-89 50 57 107 90-99 6 13 * General Practitioner, Llanidloes, Mont. 19 t Research Officer, Department of Statistccs, University College of Wales, Total 198 173 371 Aberystwyth. 14 7 October 1967 of vement-Rees and Mortality Berea, Lutkins MEDICALBRITISJOURNAL

shown in Table V, and this distribution is expressed as a the 1 % level the value of u is 2.33. For very small values of percentage of those at risk in Table VI. A and B exact values for the probabilities were obtained from TABLE IV.-Close Relatives Alive at Time of Actual Death (Survey Tables of the Hypergeometric Probability Distribution by Br Med J: first published as 10.1136/bmj.4.5570.13 on 7 October 1967. Downloaded from Group) and Hypothetical Death (Control Group) Lieberman and Owen, but none of these small sample tests were significant. Male Female Total I The statistical significance of the differences in mortality recorded between close relatives in the and control Survey Group survey Spouse 51 (5-7%) 105 (11-6%) 156 (17-3%) groups during the first year of bereavement is given in Parent 16 (1-8%) 19 (2-1%) 35 (3-9%) Table VII. Children 201 (22-3%) 235 (26-0%) 436 (48-3%) Sibling . 123 (13-6%) 153 (16-9%) 276 (30-5%) Total ..I 391 (43-3%) 512 (56-7%) 903 (100%) All Close Relatives Control Group Spouse .. .. 51(8%0/) 115 (13-1%) 166 (18-9%) The number of close relatives who died each year following Parent . .. 9 (1-0%1) 20 (2-3%/) 29 (3-3%) Child *. 222 (25-30o 188 (21-4%) 410 (46-7%) bereavement is given in Table V. This shows that there is a Sibling . .. 132 (15.00/~) 141 (16-1%) 273 (31-1%) considerable difference between the number of deaths recorded Total .. 414 (47-2%) 464 (52-8%) 878 (100%) in the survey and control groups and that the difference is

12

Statistical Method 10. LA0 BEREAVED GROUP 0 The following test was used to determine whether the pro- 8 portion of deaths in the survey group was significantly greater x than the proportion of deaths in the control group. a.0o w',,j 6 For large samples u is approximately normally distributed C with zero mean and unit variance, where

4

0 2 I CONTROL GROUP *1

0 which is algebraically equivalent to -I -2 -3 -4 YEARS OF BEREAVEMENT

FIG. 1.-Percentage of deceased people whose u=a- N death was followed each year by the death of a close relative. [ABr (N -r)]i

greatest during the first year of bereavement. During the first year 43 (4.76%) bereaved close relatives died compared with a x the notation for 2 2 and http://www.bmj.com/ using customary contingency table, 6 (0.68%) in the control group. This difference in risk a correction for continuity. between 'the two groups was found to be highly significant at A one-tail test is appropriate here, so a value of u greater the 0.001 % level. During the second year the comparative than or equal to 1.64 will be significant at the 5 % level. For figures for the bereaved and control groups were 18 (1.99%) TABLE V.-Distribution of Relatives who Died Following Bereavement

-1 Year -2 Years -I- -3 Years -4 Years -5 Years -6 Years Total Total Total Total 6S g! 6 Total g? Total on 29 September 2021 by guest. Protected copyright.

Survey Group 2 3 2 Spouse 10 9 19 2 9 3 3 2 Parent 3 2 5 2 2 2 3 1 1 Child . 6 1 7 2 1 2 Sibling 6 6 12 5 1 2 3 4 I 2 2 ITotal. . 25 18 43 6 112 18 6 4 10 4 4 8 2 2 4 - 2 2 Control Group Spouse .. 2 - 2 1 4 5 2 1 3 2 - 2 - 4 4 - - - Parent .. I.1 1 - - -- 1 1 ------Child. .. -- - 3 - 3 ------1 - 1 - - Sibling .. 1 2 3 2 1 3 - 3 3 2 1 3 1 - 1I Total. .. 4 2 6 6 5 11 2 5 7 4 1 5 2 4 6 1 - 1

TABLE VI.-Distribution of Relatives who Died Following Bereavement Expressed as a Percentage of those at Risk each Year -1 Year -2 Years -3 Years -4 Years -5 Years -6 Years 6 Total 6 Total 6 9 Total1 6 Total 6 Total 6 Total Survey Group Spouse ..19-6 8-5 12-2 4-9 7-1 6-6 7-7 - 2-5 2-8 2-2 2-4 5-7 - 1-6 - - - Pruent . ..18-8 10-5 14-3 14-3 - 6-7 9-1 11-6 10-7 - 6-6 3-6 - 7-1 4-2 - - - Child. -- 3-0 0-4 1-6 0-5 0-4 0-5 0-5 0-4 0-15 ------Sibling .. 4-8 3-9 4-3 0-9 2-7 1-9 0-9 0-7 0.8 2-4 0-7 1-6 - 0-7 0-4 - 0-7 0-4 Control Group Spouse . 3-9 - 1-2 2-0 3-5 3-0 4-2 0-9 1-9 4-3 - 1-3 - 3-6 2-6 - - - Parent. .11-0 - 3-4 - - - - 12-5 3-6 ------

------Child. -- 1-3 0-7 0-5 0-2 - - - Sibling 0-8 1-4 1-1 1-5 0-7 1-1 - 2-6 1-1 1-6 0-7 1.1 0-8 - 0-4 0-8 - 0-4 7 October 1967 Mortality of Bereavement-Rees and Lutkins MEDICAL JOURNAL 15

not TABLE VII.-Statistical Significance of the Differences in Mortality and 11 (1.25%); the difference between these figures was Recorded During the First Year of Bereavement significant. Br Med J: first published as 10.1136/bmj.4.5570.13 on 7 October 1967. Downloaded from The difference between the groups is exemplified further by Total Male Female Fig. 1; this shows the percentage of deceased people whose All relatives .. 0001 0 01 0.1 Parents .. N.S. N.S. N.S. death was followed each year by the death of a close relative. Widowed .. .. 001 5-0 10 survey group Children 5 0 5 0 N.S. It shows that close relatives of 11.6% (43) of the Siblings 5-0 N.S. N.S. died within a year of bereavement and that the corresponding All relatives (excluding widowed people). 0.1 1.0 N.S. figure for the control group was only 1.6% (6). It also shows that while the mortality rate for the control group shows a N.S. denotes that the levels were not found to be statistically significant. relatively small fluctuation over the six-year period there is an apparent change in pattern for the survey group, and the mortality rate for the bereaved group rerhains higher than that Place of Death of the control group for the first three or four years following bereavement, An assessment was made to determine whether the place at The risk for male relatives was found to be greater than which people die is related in any way to the subsequent for female relatives. During the first year of bereavement mortality of close relatives. The various places at which death 6.4% of the 391 male relatives at risk died, compared with occurred and the number of deaths which occurred at each 3.5% of the 512 female relatives. This difference in risk site are given in Table VIII. This shows that in this area, between the sexes was significant at the 5% level. though little difficulty is experienced in admitting the old, the chronic sick, and the dying into hospital, a very high propor- tion of people die at home. Because the place of death is often chosen by the patient, his family, or the people caring Widows and Widowers for him, the fact that 25.9% of the people without close relatives die at home is noteworthy. During the first two years of bereavement the mortality rate for widowed people in the bereaved group was higher than TABLE VIII.-Places at Which People Died for those in the control group. The difference between the two Chronic Old Other Total groups is shown in Tables V and VI. Of recently widowed Home Hospital Sick People's Sites people, 12.2% died during the first year of bereavement, 6.6% Unit Home of those at risk died during the second year, and 2.5% during Original Sample the third year. Though 12.2% of widowed people in the Male .. .. 108 86 24 12 14 244 bereaved group died during the first year of bereavement only Female.. .. 104 70 57 8 5 244 1.2% of the control group died. This increase in risk for Total .. .. 212 156 81 20 19- 488 % .. .. 43-4 32-0 16-6 4-1 3 9 100 widowed people during the first year of bereavement was significant at the 0.01% level. No Close Relatives Male.. .. 10 9 12 9 2 41 The mortality rate for widowers was considerably higher Female.. .. 18 12 29 6 1 67 than the mortality rate for widows. During the first year of Total.. .. 28 21 41 15 3 108 bereavement 19.6% of widowers and 8.5% of widows in the % .. .. 25-9 19-4 38 0 13-9 2-8 100 bereaved group died. This difference in risk between the Those with Close Relatives http://www.bmj.com/ widowers and widows was significant at the 5% level. Male.. .. 98 78 12 3 12 203 Of the bereaved widowers, 13.7% died during the first six Female. .. 86 57 28 2 4 177 Total .. 184 135 40| 5 16 380 months of bereavement, 5.9% died during the second six ..48-4 35-6 10-5 113 4-2 100 months, and 4.9% died during the following 12 months. % This different mortality rate for widowers during the first and second six months of bereavement was significant at the 1 % Table IX gives the percentages of people who died each year, level and confirms the findings of Young et al. (1963) that the at the sites listed, whose -death was followed by the death of a mortality rate of widowers is particularly high during the first close relative. Close relatives of 13 (7.1%) of the 184 people on 29 September 2021 by guest. Protected copyright. six months of bereavement. Conversely, the suggestion by who died at home died within the ensuing year, compared with Cox and Ford (1964) that the mortality of widows is higher the relatives of 17 (12.6%) of the 135 people who died in a than average in the second year of bereavement was, in this general hospital and 7 (17.5%) of the 40 people who died in a survey, found to be not significant at the 10% level and chronic sick unit. Close relatives of 24 (13.7%) of the 175 therefore not confirmed. people who died in hospital died within the following year. It seems, therefore, that the risk to a bereaved close relative of dying within one year of bereavement is determined, in part, by the place of death of the first relative. If the first relative Close Relatives Excluding Widowed People dies in hospital the bereaved close relatives carry twice the risk of dying within a year of bereavement than if the first Apart from widowed people, there were 747 close relatives relative had died at home. This increase in risk for bereaved in the bereaved group and 712 in the control group. During close relatives when the original death occurred in hospital the first year of bereavement 3.21% of these close relatives died compared with at home was significant at the 5% level. in the survey group compared with 0.56% in the control group. This increase in risk for bereaved relatives was significant at TABLE IX.-Percentages of People Who Died Each Year, At the Sites the 0.1% level. Listed, Whose Death was Followed by the Death of a Close Relative The mortality rates for the bereaved parents, children, and Years of Bereavement siblings are shown in Tables V and VI. These show an -1 -2 -3 -4 -5 -6 increase in mortality during the first and sometimes the second Died at home.. .. 7-1 5-0 1-2 2-5 0-6 0 6 year of bereavement for each type of relative and that in each Died in hospital .. 12-6 3-7 3-8 4 0 2-0 1.0 case the increase in mortality for the bereaved male relative is Died in chronic sick unit. 17-5 01 6-7 - greater than for the female relative. The statistical significance Died in old people's .0>0 home -37 20.0 - of the differences in mortality determined for these various Died at other sites. 37-5 - 20-0 - |_ groups is shown in Table VII. 16 7 October 1967 Mortality of Bereavement-Rees and Lutkins BRITEISCHURNAL Similarly, a difference in risk was found to occur when the significant number of relatively young people are particularly original death occurred at some site other than at home or at risk during the first year of bereavement. hospital. It was found that 19 people died at sites other Br Med J: first published as 10.1136/bmj.4.5570.13 on 7 October 1967. Downloaded from than hospital or home and that 16 of these people had close relatives living in the area. The " other sites " included roads, fields, a , chapel, shop, wood, bowling-green, and pond. Summary Deaths at these sites were invariably sudden and the shock to relatives consequently great. Their was often increased by A survey was made in a semirural area (pop. 5,184) to the need for a necropsy and . Six (37.5%) of these determine whether bereavement produced an increased mortality deaths at other sites were followed within one year by the death among bereaved close relatives-these being defined as a of a close relative, and of the close relatives at risk during this spouse, a child, a parent, or a sibling. During a six-year survey period 11.6% (6) died. period 488 residents died; 371 of these people had close Close relatives of 13.7% of the people who died in hospital relatives living in the survey area, and the subsequent mortality and of 37.5% of those who died at other sites died within one of their 903 close relatives was compared with that of a control year of bereavement. This increase in risk for the relatives group; this control group was composed of the 878 close of people who died at other sites compared with those who relatives of 371 people matched by age, sex, and marital state died in hospital was significant at the 5% level. with the people who had originally died. Close relatives of 7.1 % of the people who died at home and It was found that 4.76% of bereaved close relatives died of 37.5% of those who died at other sites died within one year within one year of bereavement compared with 0.68 % in the of bereavement. This increase in risk for the relatives of people control group. This sevenfold increase in risk between the who died at other sites compared with those who died at home bereaved and control group was significant at the 0.001%0 was significant at the 0.01% level. level, and indicates that bereavement carries a considerably increased risk of mortality. This risk was found to be greater for male than for female relatives: 6.4 % of the 391 male Age of Relatives who Die Following Bereavement relatives at risk died, compared with 3.5 % of 512 female relatives. This difference in risk between the sexes was The age distribution of all the people resident in the area significant at the 5 % level. who died during the six-year period is given in Table II. The The increase in risk was particularly great for widowed mean age of death for this group was 73.35 + 13.8. The mean people and least for female children. During the first year of age of those people whose death was followed within one year bereavement 12.2% of widowed people died, compared with by the death of a close relative was 70.9 + 13.2. The mean age 1.2% in the control group. This increase in risk for widowed of the relatives who died during the first year of their bereave- people was significant at the 0.01% level. Of the remaining close relatives in the survey group, 3.21 % died during the '21 first year of bereavement, compared with 0.56% in the control group. This increase in risk for bereaved close relatives, l0 - excluding the widowed, was significant at the 0.1 % level. B x There is a relation between the place at which a person dies and the subsequent mortality of bereaved relatives. The

risk of close relatives dying during the first year of bereave- http://www.bmj.com/

0 ment is doubled when the primary death causing bereavement occurs in a hospital compared with at home. This difference in risk was significant at the 5 % level. x 4 x-X If the primary death occurs at some site-for example, a road or field-other than at home or hospital, the risk of a z close relative dying during the first year of bereavement is the of who 2 02 x five times the risk carried by close relatives people die at home. This difference in risk was significant at the on 29 September 2021 by guest. Protected copyright. 10 20 30 40 O 70 80 90 100 AGE IN YEARS bO 0.01% level, though the small size of the sample reduces the FIG. 2.-Age distibution of relatives who died with- significance that can be attributed to the result. in one year of bereavement. People who die following a bereavement are on average slightly younger than the relatives who predeceased them, and ment was slightly lower at 69.75 + 14.7. These figures suggest they die at an earlier age than is usual for the community in that people who die following bereavement are on average which they live. slightly younger than the relatives who predeceased them and that they die at an earlier age than is usual for the community We are grateful to Dr. Felix Richards for providing data used in which they live. Fig. 2 shows the age distribution of in the survey and to Miss Kathleen Vaughan for her considerable relatives who died within one year of bereavement. This chart help in identifying the close relatives resident in the area. is biphasic, with two modes, A and B. Histograms (not shown) of the age distribution of all the people who died in the area and of those people whose death was followed within one year REFERENCES by the death of a close relative each have only one mode, and Cox, P. R., and Ford, J. R. (1964). Lancet, 1, 163. these occur in the same general age distribution as mode B in Kraus, A. S., and Lilienfeld, A. M. (1959). 7. chron. Dis., 10, 207. Fig. 2. The presence of mode A in the chart suggests that a Young, M., Benjamin, B., and Wallis, C. (1963). Lancet, 2, 454.